Search Results

The searches yielded 41,928 results, of which 24,023 were removed as duplicates. After screening titles and abstracts, we assessed 1,964 full-texts for inclusion. One hundred and nineteen met the inclusion criteria and we extracted the data of these studies. Figure 1 presents the full results of the study selection process.

After extracting data, we examined the combinations of exposure and outcomes and removed any effects that appeared more than once, keeping the effect with the largest total sample size. This process yielded 140 unique effects/outcome combinations contributed from 65 papers. The most frequently reported exposures were general tv programs and movies (n = 20), lifestyle risk behaviour (at school) intervention (n = 12), physically active video games (n = 12), and general screen use (n = 9). The most frequently reported outcomes were body composition (n = 31), general literacy (n = 11), general physical activity (n = 10), and general learning (n = 9). In most cases (120/169), there was only one exposure/outcome combination, with 17 appearing twice, and 3 three or more times. Full characteristics of the included studies are provided in Table X.

Study Quality

The quality of the included studies was mixed. Most studies assessed heterogeneity (n = 98), reported the characteristics of the included studies (n = 96), and used a comprehensive and systematic search strategy (n = 93). Most studies did not clearly report if their eligibility criteria was predefined (n unclear = 81). Many studies also did not complete dual independent screening of abstracts and full text (n = 35) or did not clearly report the method of screening (n = 36). A similar trend was observed for dual independent quality assessment (n high = 57; n unclear = 31). Overall, only 7 studies were graded as low on all criteria.

Education Outcomes

There were 43 unique effects with education outcomes, including general learning outcomes, literacy, numeracy, and science (see Figure X). Of these, 12 met our criteria for robust evidence. General screen use, TV viewing, and video games were all negatively associated with academic performance, and general screen use was further negatively associated with general learning and general literacy outcomes. Conversely, numeracy focused video games and screen-based interventions which target mathematics were positively associated with numeracy outcomes. There was substantial evidence to suggest that general screen time that occurred with an adult (i.e., coviewing) and education television are not associated with literacy outcomes (i.e., were not significantly different from zero). It is notable that of these results, heterogeneity was high for all except for the video games and academic achievement association.

Health Behaviour Outcomes

We identified 45 unique health outcome exposure combinations, which were further grouped into diet, healthy behaviour, physical activity, risky behaviour, screen time, and sleep. See Figure X for all combinations of exposure and outcome. five effects met the criteria for certainty. Advertising of unhealthy foods—both traditional advertising and video games developed by a brand for promotion—were associated higher food intake. Social media use was also associated with higher risk of substance abuse, while screen-based interventions which target health behaviours appear effective. Television was negatively correlated with sleep duration, but only at the 95% confidence level, and was not statistically significant at our more stringent 99.9% confidence level. As with the education outcomes, there was substantial unexplained heterogeneity around the findings.

Physical Health

There were 18 unique effects identified for physical health, but only five met the criteria for certainty. General screen use, television viewing, and video games were all detrimentally associated with body composition, although all also had high levels of heterogeneity. None of the effects in the other categories—cardiometabolic health, eye health, and other physical health—met our criteria for certainty.

Psychology

We identified 34 unique effects which were categorised as psychological outcomes, with six meeting the criteria for certainty. General screen use was associated with a small increased risk of depression. Social media use was associated with greater risk taking, and with higher instances of safe sex. Viewing of sexy media was associated with high sexual activity and with higher initiation of sex. All of these effects had substantial heterogeneity.

Assumptions

I made the following assumptions which may or may not be reasonable:

1. That \(\beta\) is equivilent to \(r\) for \(\beta < |.5|\) and undefined otherwise (never the case here)
2. That, given the large sample sizes involved Hedge’s g and Cohen’s d where essentially equivalent and that sample sizes used to make up these effects were approximately equal in treatment and control groups.
3. In the absence of information that could be used to construct a 2x2 contingency table that ORs and Relative Risk Ratios were equivalent and thus Relative Risk Ratios were converted to \(r\) as if they were Odds ratios.
4. External to any other information that all forms of SMD, weighted means etc. were equivilent to Cohen’s d.

Project Structure

## here() starts at C:/Users/taren/GitHub/screen_umbrella
## ! Using an auto-discovered, cached token.
##   To suppress this message, modify your code or options to clearly consent to
##   the use of a cached token.
##   See gargle's "Non-interactive auth" vignette for more details:
##   <https://gargle.r-lib.org/articles/non-interactive-auth.html>
## i The googledrive package is using a cached token for 'taren.sanders@acu.edu.au'.

The following warnings were noted during the project build:

## # A tibble: 24 x 2
##    name                  warnings                                               
##    <chr>                 <chr>                                                  
##  1 export_plots_129713ca Removed 1 rows containing missing values geom_segment. 
##  2 export_plots_1475a313 Removed 1 rows containing missing values geom_rich_tex~
##  3 export_plots_215a1c7f Removed 7 rows containing missing values geom_rich_tex~
##  4 export_plots_2a5ffd54 Removed 7 rows containing missing values geom_rich_tex~
##  5 export_plots_32a2145c Removed 2 rows containing missing values geom_rich_tex~
##  6 export_plots_3a9544fe Removed 7 rows containing missing values geom_rich_tex~
##  7 export_plots_3fd17790 Removed 2 rows containing missing values geom_rich_tex~
##  8 export_plots_56d8a19f Removed 7 rows containing missing values geom_rich_tex~
##  9 export_plots_5ab5c41d Removed 10 rows containing missing values geom_rich_te~
## 10 export_plots_609652b1 Removed 1 rows containing missing values geom_segment. 
## # ... with 14 more rows